摘要
负荷聚类特性分析是实现配电网的定制电力、高品质供电、高可靠性供电的重要基础.然而现有的Kmeans聚类分析方法,受限于数据样本集和聚类初始中心的选取等,会出现因初始中心不同造成聚类结果差异大的问题.为此,针对配电网负荷数据特点,提出一种基于改进萤火虫算法和K-means算法结合的配电网负荷聚类特性分析方法.利用萤火虫优化算法全局搜索能力强的优势,考虑类内相似度和类间差异度,寻优K-means算法初始中心,使聚类结果的聚类有效性指标取得最小值;进一步针对萤火虫算法在处理负荷数据时的弱点,通过密度法为萤火虫算法加入优秀初代个体,改进吸引公式以及个体间概率吸引移动的方式优化迭代过程中的个体移动方式,加快萤火虫算法前期收敛速度,并实现后期稳定收敛,算法更快地接近极值,计算速度更快.算例验证了本文所提算法的聚类有效性,并针对某配电台区电力负荷数据,寻得K-means算法最优初始中心,使得聚类结果的戴维森堡丁指标(Davies-Bouldin index,DBI)最小,负荷聚类结果类内差异小,类间差异大,最终聚类中心的特征代表性强,为负荷类型划分、聚类特性分析提供重要依据,为需求侧差异化电力服务定制奠定有力基础.
The load clustering analysis of distribution networks is the basis for customized power and high-quality and reliable power supply.However,the existing K-means clustering methods are limited by the data sample set and selection of the initial center of clustering,wherein selecting different initial centers will yield significantly different clustering results.Therefore,according to the characteristics of distribution network load data,a distribution network load clustering analysis method based on the combination of improved firefly and K-means algorithms is proposed.Benefiting from the strong global search ability of the firefly algorithm and considering the intraclass similarity and interclass differences,the initial center of the K-means algorithm is optimized to obtain the minimum value of the clustering effectiveness index of the results.Further,to mitigate the weakness of the firefly algorithm in processing load data by adding excellent first-generation individuals via the density method and using an improved attraction formula and a probability attraction between individuals to optimize the individual movement mode in the iterative process,the early convergence speed is accelerated,and the later convergence is stabilized.The algorithm approaches the extreme value faster with a faster calculation speed.This paper lists several examples to verify the clustering effect of the proposed algorithm and finds the optimal initial center of the K-means algorithm for the power load data of a distribution station area,thus obtaining the minimum Davies-Bouldin index,reducing the intraclass loading clustering differences and increasing the interclass differences.The final clustering center exhibits highly representative characteristics,thus providing a basis for load type division and clustering characteristic analysis and laying the foundation for differentiated power service customization on the demand side.
作者
王继东
顾志成
葛磊蛟
赵长伟
贾东强
Wang Jidong;Gu Zhicheng;Ge Leijiao;Zhao Changwei;Jia Dongqiang(Key Laboratory of Smart Grid of Ministry of Education(Tianjin University),Tianjin 300072,China;Chengdong Power Supply Branch,State Grid Tianjin Electric Power Company,Tianjin 300650,China;State Grid Beijing Electric Power Research Institute,Beijing 100075,China)
出处
《天津大学学报（自然科学与工程技术版）》
EI
CAS
CSCD
北大核心
2023年第2期137-147,共11页
Journal of Tianjin University：Science and Technology
基金
国家自然科学基金资助项目(No.51807134)
国网天津市电力公司科技资助项目(KJ21-1-18).
